Making a Par Times Chart

[Thomason]

I’m constructing a par times chart for the first time and I’ve hit a snag. (Yes, I know you can buy them but I wanted to construct one from scratch so that I would understand the process.) I chose Fair Grounds and I’ve recorded the second call and final times for every race since the current meet opened in November. I have several hundred races in the database now. (I’m working with Microsoft Access.) Last night I ran a query for all 6 furlong, open claiming races on dirt for older males. Here’s the count of the various classifications.

30K-- 1 race
30K n2L-- 2 races
25K-- 2 races
20K n2L-- 1 race
15K-- 1 race
15K n2L-- 1 race
12.5K-- 2 races
12.5K n2L-- 1 race
10K n3L-- 2 races
10K n2L-- 3 races
7.5K-- 1 race
5.5K-- 1 race
5K-- 1 race
5K n4L-- 1 race
5K n2L-- 2 races

I guess I was expecting the races to fall into neat categories or somesuch. Obviously, that hasn’t happened. So, how should I group them? Should I include the n2L, n3L races with those that don’t have the restrictions? Is my sample large enough? (I’m not sure that I know enough to ask intelligent questions.) Any assistance will be much appreciated.

[Dave Schwartz]

Thomason,

I am busy getting the 2003 HorseStreet Pars out the door Monday, but if you can give me until this time next week I could spend a little time with you on the telephone outlining what we do.

If you are interested, just visit our website and drop me an email or call. Go to the Contacts page to find me.

We are: http://www.HorseStreet.Com

Regards,
Dave Schwartz

[Thomason]

Dave,

Thank you for the very gracious offer. I'll send you an email.

Michael Thomason

[Richard]

Thom,
This may or may not help you,but I use the program EZCAPPER (Cynthia Publishing Co.)to maintain up-to-date profiles on four different tracks (AQI,PHA,SA,GG).Profiles are catagorised by following:
Track,
Distance,
Surface condition,
Gender,
Age,
Class level,
Open or statebred.
I suggest that you obtain a copy of Dick Mitchell', WINNING THOROUGHBRED STRATEGIES (currently out of print but try used book sellers at www.amazon.com).Which can explain this much better than I can in this post.

[Thomason]

Thank you, Richard. I'm not familar with EZCapper but I'll logon to Mr. Mitchell's website and take a look. I think I have the book you mentioned. Are you talking about the winner's profile, i.e., position and beaten lengths at the second and final calls? (I believe Messrs. Brohamer, Schmidt, Hambleton and others of the Sartin group were big proponents.)

I noticed your selection of tracks, two on the east coast and two on the west. I assume you're using some form of pace analysis if you're keeping track profiles. If so, I'm curious. Have you found any difference in the prices from the two coasts? I usually structure my analysis for each race using pace and while I follow the west coast tracks, I find it difficult to get a good price on a horse with a pace advantage in a California race. As a result, I usually play the tracks in the middle part of the country, e.g., Fair Grounds, Churchill Downs, Keeneland and Arlington. I've assumed that's because there are more pace handicappers on the west coast.

Michael Thomason

[GR1@HTR]

Barry Meadow had a good guest writer in his newsletter last year on the subject of Par times (i think) and home made speed figs. Very interesting...

[jotb]

Hello Thomason:

Welcome to my world! I have been working on condition pars for the NY circuit and believe me, you are in for the ride of your life. Talk about some mind bending work! This is the first of many obstacles you will run into.(not enough races per condition). My suggestion would be to fill the database with at least 5k races. Even with this large amount of races there will be many conditions that just do not make any sense. The theory of consolidating conditions is a yes and no. Yes, because there are some conditions that are basically the same on overall average. No, because many conditions are not equal in strength. It certainly would not be ethical to incorporate a certain condition that normally is faster a length on average into a group of conditions that are usually a 1/2 or 1 full length slower on average. My racing circuit has 15 racing conditions for older horses and slightly less for the 2 and 3 yo's. I demand at least a minimum of 20 races per condition for each distance for each age group as a sample. Of course this means plenty of race results especially when it comes to the longer distances. If you plan on working with a small sample for each condition, I'm certain that you will need to adjust plenty of race conditions. I remember there was a certain condition that just did not make sense. It was much slower on average than what it should have been. I had 15 races as a sample and wanted to know why this had occured. I had to go back to all my issues of simo weekly to see why. I noticed that all 15 races for this condition were run on day's where the track was much slower than normal. Kind of weird the way it happened but it did. Let's just say that the average time was 112.00 for this condition but really should be 111.00. Now this particuliar condition is run on a certain day when the track was averaging 2 lengths slower than par and the race's final time is 112.00. If the track is 2 lengths slower on average for this day, I now proceed to adjust the 112.00 to 111.60. This race now seems to look pretty good in terms of final time but if I had the true average of 111.00 and adjusted the final time of 112.00 by 2 lengths, 112.00-.40=111.60 this race would be 3 lengths slower than par and not look nearly as good as the other adjustment. The need for accurate pars is a definate but it is the amount of work you put into the condition pars that lead to the accuracy. Many par chart makers will "leave out certain conditions" because of the small sample but what happens when 2 or 3 of these "leave out conditions" run on the same card? Now, you proceed to calculate what the track variant is for the day but now are left with only 6 or 7 races to work with. What if 2 of these 6 or 7 races are run on the turf? Now you are down to 5 dirt races left to make a variant. For me that is certainly not enough races to work with to calculate a track variant. My other suggestion would be for you to TAKE YOUR TIME WHEN TRYING TO DEVISE YOUR OWN PAR CHARTS. DO NOT RUSH! A few other suggestions.

1- Collect as many race results as possible for your racing circuit.

2- If you plan on working your whole racing circuit it might be best to understand how each racing surface within your circuit differs in times from one surface to the other. This would include internal times and final times for each distance.

3- I would NOT work the 2 yo races as a whole. It might be best to break them up during the year due to their rate of muturity especially in route races. You might want to incorporate newly turned 3 yo's into the 2 yo sample. Keep the newly turned 3 yo's with late in the season 2 yo's until March or April.

4- Look for differences between gender in terms of pace and final times.

5- Do not incorporate MSW STATE BREDS into OPEN MSW'S. I know from NY that there is a major difference in pace and final times.

Good Luck with your pars,
Joe

[Thomason]

Thanks, Joe. Great post. It'll take a few days to absorb it. One point of clarification (I'm sure there will be others), you mentioned adjusting race conditions. Initially, you won't have a standard from which to work, i.e., you don't have a par times chart so you don't know when a race is fast or slow. So I assume this is a gradual process of fine-tuning. Once you've developed a rudimentary chart, you hone it with additional data.

Michael Thomason

[Richard]

Thom,
In relation to your inquiry,I find that at either SA or GG,the first couple of races on either card are often short (and also the prices).It's the back half of either card that the fields are more full.

[superfecta]

Originally posted by Thomason
I’m constructing a par times chart for the first time and I’ve hit a snag. (Yes, I know you can buy them but I wanted to construct one from scratch so that I would understand the process.) I chose Fair Grounds and I’ve recorded the second call and final times for every race since the current meet opened in November. I have several hundred races in the database now. (I’m working with Microsoft Access.) Last night I ran a query for all 6 furlong, open claiming races on dirt for older males. Here’s the count of the various classifications.

30K-- 1 race
30K n2L-- 2 races
25K-- 2 races
20K n2L-- 1 race
15K-- 1 race
15K n2L-- 1 race
12.5K-- 2 races
12.5K n2L-- 1 race
10K n3L-- 2 races
10K n2L-- 3 races
7.5K-- 1 race
5.5K-- 1 race
5K-- 1 race
5K n4L-- 1 race
5K n2L-- 2 races

I guess I was expecting the races to fall into neat categories or somesuch. Obviously, that hasn’t happened. So, how should I group them? Should I include the n2L, n3L races with those that don’t have the restrictions? Is my sample large enough? (I’m not sure that I know enough to ask intelligent questions.) Any assistance will be much appreciated. How about the times themselves listed instead of the # of races?Then group them in the same range and then check that against upcoming races.You may find you can group classes together and save alot of work,and lose very little win pct.You may also find that certain tracks have anomalies in certain distances or class levels and that can help immensely.What may be the norm 6f at FG pace wise is bad at LD,but at another distance (say 5.5f)they will equate.It shouldn't be that way but see for yourself.Hope I understand your question.

[MV McKee]

It isn't all that often that I read a string that I feel I actually have something to contribute that may benefit others. Many of the methods I use are a bit arcane. I don't want to get in the habit of posting (boasting?) about the effectiveness of one of my methods or ratings (which I may have given a special name) and yet leave the definition, logic or mechanics behind said method as a kind of nebulous alchemy or some kind of proprietary formula.
My eyes have been opened many a time reading this board and I seek to return the favor when I feel I have something of worth to others.
That being said, I wanted to pose a few questions regarding par times, and supply my answers to them. I hope it is understood that my "answers" are not the "correct" answers, only one of a myriad of possible explanations. I'll probably do this one at a time, as in most cases one of my points builds upon another, and I am not exactly a world-class typist. If there is not an audience for this stuff, please let me know. and I will cease and desist.

Point 1) If there is a winner's par for each given class, there must also be a 2nd place par, a 3rd place par, etc. In fact, in all of my studies, the pars for 2nd and 3rd place are actually more stable for these positions. By more stable I mean that the Confidence Interval for the sampling of times at a given class/finish position is smaller than that of the same sampling of winners. In fact, the 4th place position also has a smaller Confidence Interval than that of the winner, provided only field sizes >6 are used.

Point 2) Very seldom will you get a sample size at a given distance/class/surface configuration large enough to allow you to not address the lurking variable of track variant.

At this point I should backtrack a bit. (You'll excuse me if this posting is a bit disjointed, but I am also watching my now bankrupt hockey team take on the Disneyland Ducks.) I use par times for 3 primary purposes.

a) When I initially decide to begin investing at a new track/circuit, I construct a par time chart to facilitate the creation of speed figures for the past 2 years at that track. I then use these somewhat rough figures as the foundation for the more precise projected figures that I produce while actually handicapping/witnessing the races at that particular venue.

b) I use them to identify class structure "holes" that routinely occur at given tracks. I should explain this term a bit, lest I be guilty of throwing out a somewhat unknown term, sans definition. Because of a variety of factors which may vary from track to track and circuit to circuit, it is a very common occurrence to have the par time for a condition generally perceived to be "tougher" actually be slower than another condition a rung or two lower on the traditional class ladder. I wanted to give a few specific examples of this, in part to further clarify what I am attempting to explain. I also am using these examples to demonstrate why I feel it is an ill-advised practice to construct a par time chart as a neatly linear extrapolation of the few truly solid pars that one will be able to actually glean from even 3 or 4 years worth of data.
Now understand that I structure the PP's in a format that presents me with "my view" of standard DRF or Equibase data. Two of the elements in the PP's I use that differ from the standard presentation are how I represent race conditions (specifically claiming races) and how I represent the intervening period between races for a horse.
In the case of the Race Claiming conditions, I simply replace the max claiming price in a race with a number indicating the level that claiming price occupies in that particular tracks claiming hierarchy, with "1" being the claiming level bottom. For instance, a 3200 claimer and a 10000 claimer at PM would be Clm2 and Clm7 respectively. At GG it would be Clm1 and Clm6. At Hol the 3200 would be N/A and the 10000 would be Clm2...etc., etc.
In the case of time between races, I define that not only in the standard measurement of days, but also in terms of "skips", or similarly conditioned racing opportunities that this particular animal (actually his trainer) elected to pass.

Back to my original point, because of the relative dearth/plethora of racing opportunities at various class levels/distances and the somewhat flat purse structures now prevalent, it is relatively common for lower level races to actually have more earnings potential than their seemingly higher class counterparts. It is a mistake to believe that claiming price alone remains a reliable indicator of class or ability. There are a number of factors that equate to the "claiming" price of a given horse (the price an owner or trainer is willing to part with the horse for), and many of these factors have little or nothing to do with current racing ability. Earnings potential (Opportunities * Purse) are IMO a much better gauge of the strength at a given class level.
I have also seen a common occurrence in which the Winners Par at the bottom claiming level is faster than the next level, or possibly a "class hole" 2 or 3 levels above it, but the same does not apply to the pars for 2nd and 3rd place pars at the same levels. Savvy handicappers can likely rationalize the reason(s) for this.
There are actually a number of scenarios where I have seen this occur (the Win pars being out of line, but the 2nd and 3rd pars being a bit more intuitive)..Among those are N2L vs. N2X restrictions (eye-catching maiden breakers making it 2 for 2), etc.
My entire point here is that, while convenient, it is not wise to extrapolate a par for ANY level based on the apparent relationship to those immediately above and below it. Obviously, identifying these sometimes transient "class holes" can be valuable, inasmuch as it is counter-intuitive to the majority of your competition (other bettors).

c.) I use par times, or rather the distribution characteristics of the times at a given class/distance/finish position in my betting.
A couple of brief points here. I always chuckle a bit to myself when I hear the various statements about chaos/unpredictability as it relates to races which are not run to par and/or where none of the entrants has run to the par, as if this is a relatively infrequent occurrence, or a special situation which merits special handling. Think about it for a moment..upwards of @40% (50% - those races which run exactly AT the par) are, by definition, below par.
Add to that the fact that the majority of winners exhibit some degree of improvement in their winning effort as it relates to their prior effort(s), and you have to question if this truly constitutes an unusual situation.
In the race header for every race in the PP's I use, I have not only the par for the condition/distance/finish position(s), but I also have the distribution characteristics. To me this is much more important.
Now I can make wagering decisions based on a few things implied by the distribution characteristics. Examples:

I see the par for 4+Clm1 winners at PM is a 39
I also know that 68% of the time the winner will run a 37-41
I know that 95% of the time the winner will run a 35-43
I know that the par for 4th place is 34
I know that 68% of the time 4th place will run 33-35
I see the par for F3+MSW in Summer at EmD is 47
I also know that 68% of the time the winner will run a 41-53

As I hope this demonstrates, the par (average) is not the only important descriptive statistic to look at when using pars and their charts to make variants, or wagering decisions.

Which, in a circuitous manner, brings me back to Point 2 above. The vast majority of class/distance/finish position specific samples you obtain will be either too small in number or have too large a standard deviation to produce a reliable par (as measured by the confidence interval) without applying a track variant.

When I make my set of initial pars (and I don't truly enjoy this process as much as some other analysis), I have to do so in a series of passes. Pass 1, construct a par chart based on raw times, use it to apply a crude variant. Pass 2, reconstruct the par chart with the applied variant, use that chart to construct and reapply the variant. It generally takes me 3 to 5 passes before the par chart becomes stable from pass to pass. It is a bit of a pain, but as you do it, your confidence intervals tighten up considerably and the chart becomes a somewhat viable tool. In this process you have buffered the effects of the single most influential lurking variable in your large sample(s), but there are a number of others that need to be addressed if you want to utilize par times, speed figures or just about any other performance numbers you may generate.
Among these are:
The Age Dynamic

The Racing Experience Dynamic

The "Bookends" of Encapsulated or Geographically isolated race meets

Insane Racetracks that change their surface mid-meet (that was my Emerald Downs rant)

That last sentence was an indication to me that I am tiring down the stretch and need to take leave of my keyboard. But each of the factors mentioned above do need to be addressed, and how you do it can actually be a matter of personal preference. But I can share my views on that another day.
Please let me know if anyone (who has read this far) finds any value in this posting and would like to continue the dialog. I can get a bit wordy and don't want to subject anyone to the ad-nauseum treatment. Any and all feedback is welcome.

Thanks,
Michael

[Lindsay]

Michael,

This is as interesting as any post I have ever seen on a racing message board. Do I want you to continue it? Please.

[hurrikane]

Great post MV. Continue on!

[Richard]

MV,
I know I'll have to re-read your post several times to better appreciate it (I'm not the fastest learner),but I know already that you've laid doen some very serious food for thought.There can never,IMHO,be too many postings of this caliber.More is always welcome.
Richard

[Thomason]

That sound you heard was my brain being stretched. Wow.

[jotb]

Hello Michael:

I have been studying this game forever but after reading your post (with a dictionary beside me), I felt like I was back in school. I tried so hard to absorb your information because of the interest I have with speed ratings, pars, and variants but I felt like everything was going over my head. You must have had a good curve ball when you were a kid. Just kidding!

Best regards,
Joe

[MV McKee]

Lindsay, Hurrikane & Richard,

Thank you for the kind words, some of the stuff I write appears to come out of left field, and its nice to know there is at least a small audience for it.
About 4 weeks ago I posted a similar rant regarding parallel times etc., and posed a few questions afterwards. Initially I got some great, well-thought out responses to those questions, both on the board an via e-mail. Well, I took a 9 day holiday break up to the Canadian Rockies with every intent of posting a follow up sharing some additional views. But, in the intervening time period the thread began to take a general turn towards the, shall we say "non-productive" (I am being PC here). I kind of alluded to this in the first paragraph of my posting in this thread. Anyway, in light of your response(s) I'll try and do my best to keep on top of this one.

Thomason,

That is a huge compliment, thanks. My primary purpose was to stretch a brain or two, it's a healthy excercise. My brain has been stretched more than a few times by others on this board so I wanted to return the favor. I also get a great deal of benefit from the feedback I recieve. I have been using some of these methods for upwards of 10 years, and it is very helpful to hear others opinions on ideas I have that may have never been reviewed by another set of (knowledgable) eyes.

jotb,

Trust me, the problem is in my delivery, not in your understanding.
Nearly all my professional communication is directed exclusively towards a small audience of individuals in the same profession as myself. Therefore I have gotten in the habit of using a number of precise medical/technical terms and concepts as well as industry jargon without accompanying explanation. Unfortnately this habit extends beyond the workplace and I often gloss over important details necessary to most people.
Recently my wife had to drive to the Los Angeles area. I have done the Portland to LA drive so many times I could likely do it in my sleep. In response to her request for directions I just said "Get on I-5 and head south, that's about it until you get to Valencia." Those of you on the West Coast likely understand that these directions will suffice until you get to Sacramento. Needless to say, my wife was not happy when she called from Davis the next day.
Another problem I have is that I tend to construct these postings impromptu, and they usually come out somewhat disjointed, with no real logical flow of ideas. I'm trying to work on that.

[Tom]

You are a welcome addition to our community, MV.
Impressive post. I need to re-read it a few more times.
Thanks for a thought-provoking contribution.

[freeneasy]

Iwas compiling the daily track varient for a number of tracks around the country. I had all the par times for those tracks and I made my track varients in somewhat the same like manner in which you compile your pars. That is I used the times of the first 3 finishers in calculating the the daily track varient. I thought it made sense in that it compensated for those races where the winner freaked and ran off in way above normal time, or won off uncontested in above normal fashion. The 2nd and 3rd place finishers could/would bring these timeframes back into a more proper prospective and could/would give them a sharper pinnecle to balence on.

[Doug]

I am not trying to butt heads with such informative posts.

I am not much of a times oriented handicapper, nor have I ever tried to construct a par time chart.

Just looking for opinions.
.
What about the idea of just looking at times of other races for the horses entered in todays race and if there are a couple of other races on todays card that are close to todays condition.
Not the horses times, but the times of the races. Maybe get a close enough idea about what the par for todays race should be.

Maybe adjust those times somehow by the DRF variant.

Thanks for any opinions.

Doug

[jotb]

Hello Doug:

In response to your question, I would imagine this would be extremly difficult to do. You will rarely ever see the same type of race on the card unless the race is SPLIT. You might see on some days that the conditions are identical in purse structure but gender and age are different. If by chance, a race is split, say for instance a MSW. The both races run and the final times are 110.00 and 110.80. What will you compare these two races to if you have no par for the race. If you compare them to each other by just the final times, absolutely the 110.00 is faster but the question is how much faster? Now, if you have a par for this condition and let's just say, the final time par is 110.00. Now the 110.00 may not be as strong as you first thought but is the 110.80 slower and if so, how much slower? Unless you have conditions pars or track profile pars it's very unlikely that you will be able to calculate an accurate variant. JMHO

Best regards,
Joe

[JustMissed]

Doug, Why don't you just use the Pars shown as pace and speed figs. I know these are velocity figures but they work for me just as well as actual times.

I'm looking at an old pp right now and the Pars are:E1 89, E2 87, Late 75 & Speed 71. I usually look to throw out a horse that can't run within 10 speed points of par. In the particuliar race the highest back speed was 71,70,68,66,59,59,58,56,N/A. What do you know, a lot of pigs in this 7F Maden M&F 4YO.

I could easily look up the raw times for this type race at Tampa under similiar conditions but what good would that do me when trying to compare pigs.

As it turned out, the number one rated speed horse Rosmerta, did win. She was also the 2nd rated prime power horse and the 1st rated prime power horse finished second. The place horse, by the way had the second highest backspeed at 70. Don't we wish all races were this easy.

Do not mean to discourage the pursuit of pars for actual race times, but just wanted to let you know how I use the info I already have available. I'm sure that some par time figures are better and more accurate than others and worthy of consideration.

Let me know what you think.

JustMissed
:)

[Doug]

Thanks guys,

All that makes perfect sense.

Just a little too much for me. I guess that is why I prefer Pace Shape/Position handicapping. Although when I have a closing horse I like that horse to show me a much better than par last fraction.

Just Missed. As you may know I do use a service that lists pars. I have no clue how they are calculated, nor do I care, because I only use it in conjunction with the last fraction and work pretty good for Longshots.

Thanks for the info.

Doug